Proc. Natl. Acad. Sci. USAVol. 87, pp. 5454-5458, July 1990Biochemistry

Molecular cloning and neurotrophic activities of a protein withstructural similarities to nerve growth factor: Developmentaland topographical expression in the brain

(nerve growth factor family/cDNA/neurotrophic factor/hippocampal neurons/nerve growth factor receptor binding)

PATRIK ERNFORS*, CARLOS F. IBANEZ*, TED EBENDALt, LARS OLSONt, AND HAKAN PERSSON*§*Department of Medical Chemistry, Laboratory of Molecular Neurobiology, and *Department of Histology and Neurobiology, Karolinska Institute, Box 60400,S-104 01 Stockholm, Sweden; and tDepartment of Developmental Biology, Biomedical Center, Uppsala University, S-751 23 Uppsala, Sweden

Communicated by Peter Reichard, April 25, 1990 (received for review April 2, 1990)

ABSTRACT We have used a pool of degenerate oligonu-cleotides representing all possible codons in regions of homol-ogy between brain-derived neurotrophic factor (BDNF) andnerve growth factor (NGF) to prime rat hippocampal cDNAs inthe polymerase chain reaction. The amplified DNA included aproduct with significant similarity to NGF and BDNF, whichwas used to isolate a 1020-nucleotide-long cDNA from a rathippocampal library. From the nucleotide sequence, a 282-amino-acid-long protein with -45% amino acid similarity toboth pig BDNF and rat NGF was deduced. In the adult brain,the mRNA for this protein was predominantly expressed inhippocampus, where it was confined to a subset of pyramidaland granular neurons. The developmental expression in brainshowed a dear peak shortly after birth, 1 and 2 weeks earlierthan maximal expression of BDNF and NGF, respectively. Itwas also expressed in several peripheral tissues with the highestlevel in kidney. The protein, transiently expressed in COS cells,was tested on chicken embryonic neurons and readily stimu-lated fiber outgrowth from explanted Remak's ganglion and, toa lesser extent, the nodose ganglion. A weak, but consistent,fiber outgrowth response was also seen in the ciliary ganglionand in paravertebral sympathetic ganglia. Moreover, the pro-tein displaced binding of NGF to its receptor, suggesting thatit can interact with the NGF receptor. Thus, this factor,although structurally and functionally related to NGF andBDNF, has unique biological activities and represents a mem-ber of a family of neurotrophic factors that may cooperate tosupport the development and maintenance of the vertebratenervous system.

During development of the vertebrate nervous system, a vastoverproduction of neurons is compensated for by naturallyoccurring neuronal death, which is regulated by their targets(1). Within the targets, specific proteins, referred to asneurotrophic factors, are produced in limiting amounts andthe release of these proteins is believed to regulate both thetiming and the extent of innervation (2).

In the peripheral nervous system, the most well-charac-terized neurotrophic factor, nerve growth factor (NGF),supports the development of sympathetic and neural crest-derived sensory neurons, and in the adult the maintenance ofthe sympathetic nervous system is critically dependent onNGF (3, 4). In agreement with a trophic role ofNGF for adultsympathetic neurons, the levels of both NGF mRNA andprotein correlate with the density of sympathetic innervation(5, 6). NGF mRNA and protein have also been found in thebrain, with the highest levels in hippocampus and cerebralcortex, to which the major cholinergic pathways in the brainproject (7-10). Basal forebrain cholinergic neurons can be

prevented from dying after axonal transection by addition ofNGF (11-15) and they respond to NGF in vivo by a markedincrease in fiber outgrowth (16).

In addition to NGF, one other protein, termed brain-derived neurotrophic factor (BDNF), has been shown to bepresent in low amounts (17), secreted from cells (18), and tosupport survival of embryonic sensory neurons in vivo (19).In common with NGF, BDNF supports the survival of neuralcrest-derived embryonic sensory neurons in vitro, butnonoverlapping trophic activities are suggested by the findingthat BDNF also supports placode-derived neurons from thenodose ganglia and retinal ganglion cells (20, 21), which areless sensitive to NGF (22, 23). Regulation of neuronal sur-vival in vivo in the brain by BDNF has not yet beendemonstrated, although its sites of synthesis have recentlybeen mapped by in situ hybridization where a high level oflabeling was found in hippocampal neurons (24).NGF is synthesized as a preproprotein and the structure of

both the precursor and the mature protein has been deducedfrom cDNA and genomic clones (25, 26). More recently, agenomic clone has been isolated for porcine BDNF (18). Ofconsiderable interest is the finding that the mature BDNF andNGF proteins show striking amino acid similarities, suggest-ing that they are structurally related and may be members ofa family of neurotrophic factors (18).

In this study, we report on the cloning and expression of anadditional member of the NGF family.11 Due to its restrictedexpression in the brain, being mostly confined to a subset ofpyramidal and granular neurons in the hippocampus, we havenamed this protein hippocampus-derived neurotrophic factor(HDNF).

MATERIALS AND METHODSRNA Preparation, Molecular Cloning, and DNA Sequenc-

ing. Polyadenylylated RNA [poly(A)+] was prepared as de-scribed (27). For cloning, rat hippocampus poly(A)+ RNA (5jug) was used as a template for synthesis of single-strandedcDNA using Moloney murine leukemia virus reverse tran-scriptase (Pharmacia). Six separate mixtures of 28-mer oli-gonucleotides representing all possible codons correspond-ing to the amino acid sequence KQYFYET (5'-oligonucleo-tide) and WRFIRID (3'-oligonucleotide) were synthesized onan Applied Biosystems A381 DNA synthesizer. The 5'-oligonucleotide contained a synthetic EcoRI site and the3'-oligonucleotide contained a synthetic HindIII site. Eachmixture of oligonucleotides was then used to prime theamplification of hippocampal cDNA (25 ng) by the polymer-

Abbreviations: NGF, nerve growth factor; BDNF, brain-derivedneurotrophic factor; HDNF, hippocampus-derived neurotrophic fac-tor; PCR, polymerase chain reaction.§To whom reprint requests should be addressed.$The sequence reported in this paper has been deposited in theGenBank data base (accession no. M34643).

5454

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Oct

ober

20,

202

0

Proc. Natl. Acad. Sci. USA 87 (1990) 5455

ase chain reaction (PCR) (Gene Amp, Perkin-Elmer/Cetus).PCR products of the expected size [182 base pairs (bp)including primer and restriction site] were isolated on anagarose gel and cloned in plasmid Bluescript KS' (Strata-gene) followed by nucleotide sequence analysis by thedideoxynucleotide chain-termination method (28). The insertfrom one clone showing =60% nucleotide sequence similarityto both rat NGF and pig BDNF was then used to screen acDNA library in Agtl0 from rat hippocampus constructedwith a cDNA synthesis kit (Pharmacia). Screening of 1.2 X106 independent cDNA clones from the primary library (29)yielded seven positive clones. One clone with a 1020-bp-longinsert was sequenced in its entirety on both strands by thechain-termination method.RNA Blot Analysis. Poly(A)+ RNA (20 pg) was electro-

phoresed in a 1% agarose gel containing 0.7% formaldehydeand was transferred to a nitrocellulose filter. The filter washybridized to a 355-bp fragment from the 3' end of HDNFmRNA (nucleotides 665-1020 in Fig. 1), isolated by PCRusing one specific primer together with an oligo(dT) primer asdescribed by Frohman et a!. (30). The fragment was labeledwith [a-32P]dCTP by nick-translation to a specific activity of-5 x 108 cpm/tug. Hybridization was carried out as de-scribed (27) followed by washing at high stringency andexposure to Kodak XAR-5 films. The same filters were boiledfor 5 min in 1% glycerol, followed by hybridization first to a185-bp PCR fragment from rat BDNF corresponding to aminoacids 183-239 of pig BDNF (18), then to a 770-bp BstE2/PstI fragment from the 3' exon of rat NGF (31), and finally to a1.5-kilobase (kb) Pst I fragment from a mouse a-actin cDNA(32). Appropriate exposures of all autoradiograms were quan-tified with a Shimadzu CS-9000 densitometer.

In Situ Hybridization. Cryostat sections (14 ,um) fromfresh-frozen adult Sprague-Dawley rat brain were processedand used for in situ hybridization with deoxyadenosine[a-[35S]thio]triphosphate 3'-end-labeled probes as described(33). To detect HDNF-specific mRNA, a 50-mer oligonucle-otide complementary to nucleotides 667-717 in Fig. 1 wasused. For BDNF-specific mRNA, a 50-mer oligonucleotidecomplementary to rat BDNF mRNA, corresponding to nu-cleotides 748-798 in pig BDNF (18), was used.

Expression of HDNF Protein, Assays of Biological Activities,and Binding to the NGF Receptor. The 1020-bp HDNF cDNAinsert in AgtlO was amplified by PCR using AgtlO sequencingprimer and reverse primer (Clontech). The amplified DNAwas treated with T4 polynucleotide kinase and 10-mer Xho Ilinkers were ligated, followed by cloning of the fragment intothe Xho I site of pXM (34). COS cells grown to =70%confluence were transfected with 20 ,g of the indicatedplasmid construct per 10-cm dish by the DEAE dextran-chloroquine method (35). A plasmid expressing the ,B-galactosidase gene (pCH11O, Pharmacia) was transfected inparallel and f-galactosidase activity was measured in cyto-plasmic extracts as a control of transfection efficiency. Theconditioned medium from transfected cells (36) was thencollected and assayed for stimulation of neurite outgrowthfrom embryonic chicken ganglia as described (37). Condi-tioned medium from transfected cells was tested for bindingto the NGF receptor on PC12 cells as described (36).

RESULTSMolecular Cloning and Structure of HDNF. A 1020-bp

cDNA clone was isolated. Nucleotide sequence analysis ofthis clone showed an open reading frame encoding a 282-amino-acid-long protein (Fig. 1). The C-terminal part of thisprotein contained a potential cleavage site for a 119-aminoacid protein with 57% amino acid similarity to both rat matureNGF and pig mature BDNF. Included in this similarity wereall six cysteine residues, involved in formation of disulfidebridges, and an overall identity to NGF and BDNF of 68 and

1 V D V P G N S H T D A M V T S A T I L Q 201 GTCGACGTCCCTGGAAATAGTCATACGGATGCCATGGTTACTTCTGCCACGATCTTACAG 60

21 IV N K V M S I L F Y V I F L A Y L R G I 4061 IGTGAACAAGGTGATGTCCATCTTGTTATGTGATATTTCTTGCTTATCTCCGTGGCATC 120

41 Q G N N M D Q R S L P E D S L N S L I I121 CAAGGCAACAACATGGATCAAAGGAGTTTGCCAGAAGACTCTCTCAATTCCCTCATTATC

60180

61 K L I Q A D I L K N K L S K Q M V D V K 80181 AAGTTGATCCAGGCGGATATCTTGAAAAACAAGCTCTCCAAGCAGATGGTAGATGTTAAG 2 40

81241

101301

E N Y Q S T L P K A E A P R E P E Q G E 100GAAAATTACCAGAGCACCCTGCCCAAAGCAGAGGCACCCAGAGAACCAGAGCAGGGAGAG 300

A T R S E F Q P M I A T D T E L L R Q Q 120GCCACCAGGTCAGAATTCCAGCCGATGATTGCAACAGACACAGAACTACTACGGCAACAG 360

121 R R Y N S P R V L L S D S T P L E P P P 140361 AGACGCTACAATTCACCCCGGGTCCTGCTGAGTGACAGCACCCCTTTGGAGCCCCCTCCC 420

141 L Y L M E D Y V G N P V V T N R T S P R 160421 TTATATCTAATGGAAGATTATGTGGGCAACCCGGTGGTAACCAATAGAACATCACCACGG 480

161 R K Rf Y A E H K S H R G E Y S V C D S E 180481 AGGAAACGCTATGCAGAGCATAAGAGTCACCGAGGAGAGTACTCAGTGTGTGACAGTGAG 540

181541

201601

S L W V T D K S S A I D I R G H Q V T VAGCCTGTGGGTGACCGACAAGTCCTCAGCCATTGACATTCGGGGACACCAGGTTACAGTG

L G E I K T G N S P V K Q Y F Y E T R CTTGGGAGAGATCAAAACCGGCAACTCTCCTGTGAAACAATATTTTTATGAAACGAGGTGT

200600

220660

221 K E A R P V K N G C R G I D D K H W N S 240661 AAAGAAGCCAGGCCAGTCAAAAACGGTTGCAGGGGGATTGATGACAAACACTGGAACTCT 720

241 Q C K T S Q T Y V R A L T S E N N K L V 260721 CAGTGCAAAACGTCGCAAACCTACGTCCGAGCACTGACTTCAGAAAACAACAAACTCGTA 780

261 G W R W I R I D T S C V C A L S R K I G 280781 GGCTGGCGCTGGATACGAATAGACACTTCCTGTGTGTGTGCCTTGTCAAGAAAAATCGGA 840

281 R T End 282841 AGAACATGAATTGGCATCTGTCCCCACATATAAATTATTACTTTAAATTATATGATATGC 900

901 ATGTAGCATATAAATGTTTATATTGTTTTTATATATTATAAGTTGACCTTTATTTATTAA 960961 ACTTCAGCAACCCTTACAGIEOZCTTTTTTCATAATCGGGCTGCTCAAAAAAAAAA 1020

FIG. 1. Nucleotide sequence and deduced amino acid sequenceof rat HDNF. Arrow indicates the presumptive start of matureHDNF. A consensus sequence for N-glycosylation is underlined. Aconsensus sequence for polyadenylylation is shown in the box, andthe stretches of adenosine at the end of the sequence show thepoly(A) tail. Vertical bar shows an exon/intron boundary present inthe rat NGF gene (31). Stars indicate potential translation start sites.

67 residues, respectively, was seen. The prepro part of theprotein showed weak, but significant, homology to NGF. Apotential N-glycosylation site located nine amino acids fromthe start of the mature protein was also conserved betweenthe three proteins.

Expression ofHDNF mRNA in Peripheral Tissues. A 1.3-kbHDNF-specific mRNA was detected in several adult ratperipheral tissues, with the highest level in kidney (data notshown). Densitometer scanning of autoradiograms fromthree independent experiments showed that the spleen andheart contained approximately 6- and 8-fold lower levels,respectively, compared with kidney. Lower levels were alsofound in the adrenal gland, ovary, muscle, and liver. Hybrid-ization of the same filter to a rat NGF probe revealed that theamounts of HDNF mRNA in heart and spleen were compa-rable to the level of NGF mRNA in these tissues.

Developmental Expression of HDNF and BDNF in RatBrain. In the developing brain, HDNF mRNA was detectedalready at embryonic day 15, the earliest time point tested(Fig. 2a). A sharp increase was seen at birth and maximallevels were found at postnatal day 4. At 3 weeks of age, theamount had decreased to adult levels. Densitometer scanningof autoradiograms from two independent experimentsshowed that adult levels were 15-fold lower than at postnatalday 4. The level in the adult brain was lower than in kidneybut was comparable to the level in heart. A 1.4-kb BDNFmRNA was first seen at embryonic day 19, with a peak levelat 2 weeks of age, which was 10-fold higher than the amountin adult brain (Fig. 2a). A 4.0-kb BDNF mRNA was also seenand the developmental and regional expression of this mRNAwas the same as for the 1.4-kb mRNA.

Regional Distribution of HDNF and BDNF mRNA in AdultRat Brain. The distribution of HDNF mRNA in the adult

Biochemistry: Ernfors et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

20,

202

0

Proc. Natl. Acad. Sci. USA 87 (1990)

a In~~~~~~~~~~~~~Er(onY Y)UJ N-J -Y-UJ-C

LU LU LU LUJ a. Cl) LO)

HDNF

1.3kb -

co

o _

BDNFE

1.4kb -

b-C

o0

O CDo E

HDNF

1.3kb- .

BDNF

1.4kb -

C:

-o

~0

E)

Cl)

E0 CZ

>C

xO oQ0.- E

-r o Q

O E

n X

O UW

E

a)

:iw

FIG. 2. Developmental and regional expression of HDNF andBDNFmRNA in rat brain. (a) Poly(A)+ RNA (20 pg per slot) isolatedfrom Sprague-Dawley rat brain at the indicated developmentalstages was hybridized to the indicated probes (HDNF and BDNF).Adult rats were 12 weeks old. E, embryonic day; P, postnatal day;wks, weeks. (b) Same analysis as in a using poly(A)+ RNA (20 ug perslot) isolated from the indicated regions of adult male Sprague-Dawley rat brain. Medulla, medulla oblongata; hypo, hypothalamus;hippo, hippocampus; cortex, cerebral cortex; cblm, cerebellum; olf,olfactory bulb.

brain showed remarkable regional specificity with high levelsin hippocampus compared with other brain regions analyzed(Fig. 2b). In fact, cerebellum was the only other region whereHDNF mRNA was clearly detected, with the exception of

cerebral cortex, which showed a weak signal. BDNF mRNAwas more widely distributed in rat brain, although hippocam-pus also contained the highest amount, followed by cerebralcortex, pons, and cerebellum (Fig. 2b).Neurons Expressing HDNF and BDNF mRNA Are Located

in a Distinct Topographical Arrangement in Hippocampus.Anterior sections ofthe dorsal hippocampus showed neuronsexpressing high levels ofHDNF mRNA primarily confined tothe medial part of CA1 and CA2 (Fig. 3 a and c). Few HDNFmRNA-expressing neurons were also found in lateral parts ofCAL. Granular cells of the dentate gyrus were also highlylabeled (Fig. 3a). CA3 and hilar cells of the dentate gyrmsshowed no labeling for HDNF mRNA at any level (Fig. 3d).No labeling was seen over any sections after hybridization toa control probe, complementary to the specific HDNF probe.Adjacent sections hybridized to a BDNF-specific proberevealed labeling over granular neurons in the dentate gyrus(Fig. 3b), although possibly with lower intensity than thatseen after hybridization for HDNF mRNA. Strong labelingwith the BDNF-specific probe was found over neurons in thehilar region (Fig. 3e), CA3, and part of CA2 (Fig. 3b). FewBDNF mRNA-expressing neurons, which appeared to beless intensively labeled, were also detected in CA1 and CA2(Fig. 3b). Intensely labeled neurons were seen in claustrum,located lateral to the external capsule. This region showed nolabeling for HDNF mRNA.

Neurotrophic Activities of HDNF in Explanted ChickenEmbryonic Ganglia. The 1020-bp HDNF cDNA insert wascloned in the expression vector pXM (34), designed fortransient expression in COS cells. Two plasmid constructswere isolated, containing the HDNF insert either in thecorrect or opposite orientation for translation of the HDNFprotein. The latter construct was used as a negative control.Included was also a construct containing the rat NGF gene(36). The different constructs were transfected into COS cellsand 3 days later conditioned medium was tested for biologicalactivity in bioassays that measured fiber outgrowth fromvarious chicken embryo ganglia. A marked stimulation ofneurite outgrowth, consistently resulting in circular or ovalfiber halos, was seen in the ganglion ofRemak, a ganglionatednerve trunk in the mesorectum ofthe chicken embryo (38, 39)(Fig. 4a). Although NGF is known to stimulate the explantedganglion of Remak (39), it was far less efficient than HDNF(Fig. 4b). A modest stimulation of fiber outgrowth was alsoseen with HDNF in the nodose ganglion, consisting ofneurons exclusively derived from an epidermal placode (22)

c y d. a

.,

f f

os *' *-

*' b; .' ') ;,* , .)i, , i ' *,, .., _

r f "'1.; tv * ! ,:,'' - ,'*;. + ''.' t-,

. ., g

w

j41

FIG. 3. Expression of HDNF andBDNF mRNA in hippocampal neurons.Rat (Sprague-Dawley) brain sections hy-bridized to either HDNF- or BDNF-specific oligonucleotide probes. (a) Au-toradiogram from a section at the level ofhippocampus hybridized to the HDNF-specific probe. Note labeling over medialCA1, CA2, and the dentate gyrus. (b)Adjacent section hybridized to a BDNF-specific probe. Note labeling over CA2and CA3 as well as hilar cells and dentategranule layer. (c) Pyramidal neurons inmedial CA1 labeled with the HDNF-specific probe. (d) Nonlabeled hilar neu-rons after hybridization to the HDNF-specific probe. (e) Hilar neurons labeledwith the BDNF-specific probe. DG, den-tate gyrus; CAlm, CA1 medial; Hi, hilusof dentate gyrus; Cl, claustrum. (a and b,bar = 1.3 mm; c-e, bar = 10 ,um.)

5456 Biochemistry: Ernfors et al.

to v IW.,

e-.

-1

. I .1

Dow

nloa

ded

by g

uest

on

Oct

ober

20,

202

0

Proc. Natl. Acad. Sci. USA 87 (1990) 5457

FIG. 4. Stimulation of fiber outgrowth from chicken embryonicganglia. Biological activity of recombinant HDNF shown as effectson different nerve tissues from the chicken embryo. Remak ganglionstimulated by HDNF (a) or NGF (b). (c) Nodose ganglion withHDNF. Paravertebral sympathetic ganglion in response to HDNF (d)and recombinant rat NGF (e). (I) Ciliary ganglion with HDNF. Allfigures show ganglia after 1.5 days in culture. Dark-field microscopy.(Bars = 0.3 mm.)

(Fig. 4c). Again, HDNF was superior to NGF in evoking thisresponse. A weak, but consistent, fiber outgrowth responsewith HDNF was seen in paravertebral sympathetic trunkganglia (Fig. 4d), which, however, was much less pronouncedcompared with the massive response to rat NGF (Fig. 4e). Inthe ciliary ganglion, a weak but consistent fiber outgrowthresponse, manifested by the projection of short neuritefascicles, was seen with HDNF but never with NGF (Fig. 4c).In the dorsal root ganglia, HDNF stimulated neurite out-growth to the same extent as NGF.

Displacement of NGF Binding to PC12 Cells by HDNF.Concentrated conditioned medium from transfected COScells was tested for its ability to compete for binding 125I-labeled NGF (125I-NGF) to its receptor on PC12 cells. Theconcentration of 1251-NGF used allowed "80% of the labeledNGF to be bound to the low-affinity receptor site in theabsence of competition (40). Twenty-five times concentratedmedium containing the HDNF protein displaced -70% of thelabeled NGF and a 20% displacement was seen after a 25-folddilution (Fig. 5). In contrast, 25 times concentrated mediumfrom COS cells transfected with the HDNF cDNA in theopposite orientation did not show any displacement. Con-

100-0)

c

U-

.

L(

LO

75

50

25-

0-

1:25 1:5 1:1 5:1 25:1Concentration of conditioned medium

FIG. 5. Displacement of 1251-NGF from its receptor on PC12 cellsby HDNF and NGF. Serial dilutions of transfected COS cell mediumwith (z) or without (a) HDNF or containing rat NGF (m) were

assayed for their ability to displace 1251-NGF from its receptor onPC12 cells. Data are from two independent experiments that showeda variation of ±20%.

centrated medium from cells transfected in parallel with a ratNGF gene displaced 50% of the labeled NGF when diluted250 times.

DISCUSSIONThe cDNA clone isolated in this study encodes a protein,HDNF, with a remarkable sequence similarity to both NGFand BDNF and therefore represents an additional member ofa family of neurotrophic proteins. Recently (at the time ofsubmission of this manuscript), two groups (41, 42) indepen-dent ofus isolated genomic clones for a protein (neurotrophin3) from mouse and rat, respectively, which is identical to theneurotrophic protein characterized in this study. Our cDNAclone predicts a 282-amino-acid-long protein, which is 24amino acids longer than the protein deduced from the ge-nomic clones (41, 42). Two alternative start sites for trans-lation of the NGF protein have been proposed; the first islocated in a separate 5' exon (43). The second start site,located in the 3' exon, is also efficiently used for translationof the NGF protein (36, 44) and generates a 68-amino acidshorter protein. Thus, the structure of our cDNA cloneindicates that the HDNF protein utilizes two alternative startsites for translation, located in separate exons, and suggeststhat the genomic organization of HDNF and NGF is verysimilar.

In peripheral ganglia bioassays, HDNF showed neuro-trophic activities that were to some extent reminiscent ofboth NGF and BDNF. Thus, in similarity to BDNF (20),HDNF stimulated fiber outgrowth from the nodose gangliaand, as for NGF, evoked a fiber outgrowth response insympathetic ganglia. In the latter case, however, the re-sponse was clearly weaker than with NGF. The partiallyoverlapping activities seen in vitro may reflect a cooperationof these factors in vivo, where two or more proteins from thesame family may support the development and/or mainte-nance of specific neurons. The most striking stimulation offiber outgrowth evoked by HDNF was seen in the peripheral,autonomic, ganglion of Remak containing mostly cholinergicbut also some adrenergic neurons (38, 39). This effect wasclearly more pronounced than effects seen with NGF (39),suggesting that HDNF also evokes trophic responses differ-ent from both NGF and BDNF. In agreement with this,HDNF showed a weak, but consistent, neurite outgrowthresponse in the ciliary ganglion, which does not respond toNGF or BDNF. The ciliary ganglion is known to respond tociliary neurotrophic factor (45), which lacks a signal se-quence, but could be released by an as yet unknown mech-anism (46). Thus, HDNF is the only secreted neurotrophicfactor today that is known to affect fiber outgrowth, at leastin vitro, from the ciliary ganglion.The HDNF protein displaced 125I-NGF from PC12 cells,

indicating that it can interact with the NGF receptor. With theassumption that NGF and HDNF were produced in equalamounts in parallel transfections and that the conditionedmedium lacks interfering substances, the interaction ofNGFto its receptor was 30-fold more efficient. PC12 cells haveboth low- and high-affinity receptors but only the high-affinity receptor mediates a biological response (47). The factthat recombinant rat NGF readily stimulated neurite out-growth from PC12 cells, whereas HDNF, even at 30-foldhigher concentrations than NGF, did not suggests thatHDNF can only interact with the NGF receptor in itslow-affinity form. It therefore appears likely that the biolog-ical responses elicited by HDNF are mediated by either aseparate second messenger system compared with NGF orthat the HDNF receptor is different from the NGF receptor.

In similarity with NGF, HDNF mRNA was found inseveral peripheral rat tissues, with the highest level in kidney.Hybridization of the same filters to a rat NGF probe revealedthat the level of HDNF mRNA in kidney was only slightly

Biochemistry: Ernfors et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

20,

202

0

Proc. Natl. Acad. Sci. USA 87 (1990)

higher than the levels of NGF mRNA in peripheral sympa-thetic target tissues, indicating that HDNF is produced inrelatively small amounts in peripheral rat tissues. This is alsotrue for the brain, and the fact that seven positive cDNAclones were isolated from 1.2 x 106 independent clonessuggests that in hippocampus, containing the highest level ofHDNF mRNA, this transcript constitutes =1 in every170,000, which clearly represents a rare transcript. Thus, asin the case of NGF, HDNF may be present in limitingamounts and functions in vivo as a target-derived factor fora specific subset of both peripheral and central neurons. Theregional distribution of HDNF mRNA in the periphery is,however, different from NGF, and, in agreement with the invitro biological assays, HDNF may support a different set ofperipheral neurons. Of interest is also that HDNF mRNAwas found in the ovary, whereas no mRNA was detected inthe testis, where both NGF and its receptor is expressed (48)and where NGF has been suggested to mediate an interactionbetween Sertoli cells and germ cells (49). This shows thatdifferent members of the NGF family are expressed indifferent reproductive tissues and suggests that they mayhave nonoverlapping functions outside the nervous system.

Interestingly, the three neurotrophic proteins were maxi-mally expressed at different times of brain development witha peak of HDNF mRNA shortly after birth, BDNF mRNAaround 2 weeks, and NGF mRNA around 3 weeks after birth(see ref. 8 for NGF). Moreover, the mRNA's for all threeproteins were expressed in hippocampus at levels higher thanin other regions, particularly in the case of HDNF. Withinhippocampus, all three mRNAs were also confined to neu-rons (see ref. 10 for NGF) and a clear topographical divisionwas seen, where HDNF mRNA was concentrated to pyra-midal neurons in medial CA1, CA2, and granular neurons indentate gyrus. Strongly labeled BDNF neurons were primar-ily seen in CA3 and the hilar region of dentate gyrus. Neuronswith apparent lower levels of BDNF mRNA were seen in thedentate gyrus. The hilar region, containing neurons with highlevels of BDNF mRNA, showed no labeling for HDNFmRNA.

This remarkable concentration of trophic factors in theadult hippocampus suggests that maintenance of plasticity iscrucial to its function and may relate to the presumedmorphological sequelae of long-term potentiation and mem-ory consolidation processes. The intriguing temporal andspatial expression of the three neurotrophic proteins in thebrain suggests that they predominantly support neuronalinnervation at different times of development and that theymay also exert specific trophic support for different centralnervous system neurons, a possibility that will be an inter-esting topic for future studies.

We thank Annika Jordell-Kylberg for expert technical assistancewith the bioassays. This work was supported by The SwedishNatural Science Research Council, the Swedish Medical ResearchCouncil, the Swedish Natural Environment Protection Board, Mag-nus Bergvalls Stiftelse, US Grants (University of Colorado) AG04418 and NS-09199, and funds from the Karolinska Institute. P.E.was supported by the Swedish Medical Research Council.

1. Cowan, W. M., Fawcett, J. W., O'Leary, D. D. M. & Stanfield, B. B.(1984) Science 225, 1258-1265.

2. Black, I. B. (1986) Proc. Natl. Acad. Sci. USA 83, 8249-8252.3. Levi-Montalcini, R. (1987) Science 237, 1154-1162.4. Thoenen, H. & Barde, Y.-A. (1980) Physiol. Rev. 60, 1284-1335.5. Heumann, R., Korsching, S., Scott, J. & Thoenen, H. (1984) EMBO J.

3, 3183-3189.6. Shelton, D. L. & Reichardt, L. F. (1984) Proc. Natl. Acad. Sci. USA 81,

7951-7955.

7. Korsching, S., Auberger, G., Heumann, R., Scott, J. & Thoenen, H.(1985) EMBO J. 4, 1389-1393.

8. Whittemore, S. R., Ebendal, T., LIArkfors, L., Olson, L., Seiger, A.,Stromberg, I. & Persson, H. (1986) Proc. Natl. Acad. Sci. USA 83,817-821.

9. Shelton, D. L. & Reichardt, L. F. (1986) Proc. Natl. Acad. Sci. USA 83,2714-2718.

10. Ayer-LeLievre, C., Olson, L., Ebendal, T., Seiger, A. & Persson, H.(1988) Science 240, 1339-1341.

11. Hefti, F. (1986) J. Neurosci. 6, 2155-2162.12. Williams, L. R., Varon, S., Peterson, G. M., Wictorin, K., Fischer, W.,

Bjorklund, A. & Gage, F. H. (1986) Proc. Natl. Acad. Sci. USA 83,9231-9235.

13. Kromer, L. F. (1987) Science 235, 214-216.14. Rosenberg, M. B., Friedmann, T., Robertson, R. C., Tuszynski, M.,

Wolff, J. A., Breakefield, X. 0. & Gage, F. (1988) Science 242, 1575-1578.

15. Stromberg, I., Wetmore, C. J., Ebendal, T., Ernfors, P., Persson, H. &Olson, L. (1990) J. Neurosci. Res. 25, 405-411.

16. Ernfors, P., Ebendal, T., Olson, L., Mouton, P., Str6mberg, I. &Persson, H. (1989) Proc. Nat!. Acad. Sci. USA 86, 4756-4760.

17. Barde, Y.-A., Edgar, D. & Thoenen, H. (1982) EMBO J. 1, 549-553.18. Leibrock, J., Lottspeich, F., Hohn, A., Hofer, M., Hengerer, B.,

Masiakowski, P., Thoenen, H. & Barde, Y.-A. (1989) Nature (London)341, 149-152.

19. Hofer, M. M. & Barde, Y.-A. (1988) Nature (London) 331, 261-262.20. Lindsay, R. M., Thoenen, H. & Barde, Y.-A. (1985) Dev. Biol. 112,

319-328.21. Johnson, J. E., Barde, Y.-A., Schwab, M. & Thoenen, H. (1986) J.

Neurosci. 6, 3031-3038.22. Hedlund, K. & Ebendal, T. (1980) J. Neurocytol. 9, 665-682.23. Davies, A. M., Thoenen, H. & Barde, Y.-A. (1986) J. Neurosci. 6,

1897-1904.24. Wetmore, C., Ernfors, P., Persson, H. & Olson, L. (1990) Exp. Neurol.

109, in press.25. Scott, J., Selby, M., Urdea, M., Quiroga, M., Bell, G. I. & Rutter, W. J.

(1983) Nature (London) 302, 538-540.26. Ullrich, A., Gray, A., Berman, C. & Dull, T. J. (1983) Nature (London)

303, 821-825.27. Ernfors, P., Hallbd6k, F., Ebendal, T., Shooter, E. M., Radeke, M. J.,

Misko, T. P. & Persson, H. (1988) Neuron 1, 983-996.28. Sanger, F., Nicklen, S. & Coulson, A. (1977) Proc. Nat!. Acad. Sci. USA

74, 5463-5467.29. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) Molecular Cloning:A

Laboratory Manual (Cold Spring Harbor Lab., Cold Spring Harbor, NY).30. Frohman, M. A., Dush, M. K. & Martin, G. P. (1988) Proc. Nat!. Acad.

Sci. USA 85, 8998-9002.31. Whittemore, S. R., Friedman, P. L., Larhammar, D., Persson, H.,

Gonzalez-Carvajal, M. & Holets, V. R. (1988) J. Neurosci. Res. 20,403-410.

32. Minty, A. J., Caravatti, M., Benoit, R., Cohen, A., Daubas, P., Weydert,A., Gros, F. & Buckingham, M. E. (1981) J. Biol. Chem. 286,1008-1014.

33. Ernfors, P., Henschen, A., Olson, L. & Persson, H. (1989) Neuron 2,1605-1613.

34. Yang, Y. C., Ciarlette, A. B., Temple, P. A., Chung, M. P., Kovacic, S.,Wilek-Giannotti, J. S., Leary, N. C., Kriz, R., Donahue, R. E., Wong,G. G. & Clark, S. C. (1986) Cell 47, 3-10.

35. Luthman, H. & Magnusson, G. (1983) Nucleic Acids Res. 11, 1295-1305.36. Ibanez, C., Hallbo6k, F., Ebendal, T. & Persson, H. (1990) EMBO J. 9,

1477-1483.37. Ebendal, T. (1989) IBRO Handb. Ser. 12, 81-93.38. Le Douarin, N. M., Teillet, M. A., Ziller, C. & Smith, J. (1978) Proc.

Natl. Acad. Sci. USA 75, 2030-2034.39. Ebendal, T. (1979) Dev. Biol. 72, 276-290.40. Sutter, A., Riopelle, R. J., Harris-Warrick, R. M. & Shooter, D. M.

(1979) J. Biol. Chem. 254, 5972-5982.41. Hohn, A., Leibrock, J., Bailey, K. & Barde, Y.-A. (1990) Nature

(London) 344, 339-341.42. Mainsonpierre, P. C., Belluscio, L., Squinto, S., Ip, N. Y., Furth,

M. E., Lindsay, R. M. & Yancopoulos, G. D. (1990) Science 247,1446-1451.

43. Selby, M. J., Edwards, R., Sharp, F. & Rutter, W. J. (1987) Mol. CellBiol. 7, 3057-3064.

44. Hallbook, F., Ebendal, T. & Persson, H. (1988) Mol. Cell. Biol. 8,452-456.

45. Adler, R., Landa, K., Manithorpe, M. & Varon, S. (1979) Science 204,1434-1436.

46. Stockli, K. A., Lottspeich, F., Sendiner, M., Maslakowski, P., Carroll,P., Gotz, R., Lindholm, D. & Thoenen, H. (1989) Nature (London) 342,920-923.

47. Eveleth, D. D. & Bradshaw, R. A. (1988) Neuron 1, 929-936.48. Ayer-LeLidvre, C., Olson, L., Ebendal, T., Hallb6ok, F. & Persson, H.

(1988) Proc. Nat!. Acad. Sci. USA 85, 2628-2632.49. Persson, H., Ayer-LeLidvre, C., Sbder, O., Villar, M. J., Metsis, M.,

Olson, L., Ritzen, M. & Hnkfelt, T. (1990) Science 247, 704-707.

5458 Biochemistry: Ernfors et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

20,

202

0